[
https://issues.apache.org/jira/browse/PHOENIX-4964?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16645604#comment-16645604
]
Lars Hofhansl commented on PHOENIX-4964:
----------------------------------------
It's not doing this in either case. (Albeit the optimization is rejected at
different points in the code.)
Another optimization missing is to use the index for ORDER BY ... DESC. That
could be translated to an HBase reverse scan - just as we do for a descending
ORDER BY of the pk.
> ORDER BY does not use an index if the query is not fully covered
> ----------------------------------------------------------------
>
> Key: PHOENIX-4964
> URL: https://issues.apache.org/jira/browse/PHOENIX-4964
> Project: Phoenix
> Issue Type: Improvement
> Reporter: Lars Hofhansl
> Priority: Minor
>
> I just noticed that a query like
> {{SELECT <column1> FROM <table> ORDER BY <column2> LIMIT <n>}}
> Does not use an index on <column2> if it does not also include <column1>.
> That seems too limited.
> I can see the code in QueryOptimizer.addPlan that rewrites the plan only when
> there's a WHERE clause, and then only for the WHERE clause.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
